With free software you can validate the code quality and tweak and share code as you wish. With AI slop becoming more prevalent this means choosing quality software is much easier. This is especially true given that many proprietary software companies are switching to AI which will harm them in the long run.
In short, free software is human focused. It doesn’t have economic incentives for the most part so it will always be as good as the community around it.
Except for all the script kiddies vibe-coding FOSS apps as well. The problem with the whole “validate the code quality” concept is that the potential user needs to know how to do so. The majority of users don’t.
I mean, there were plenty of shitty quality FOSS projects before AI, some of them even got popular. AI doesn’t have a monopoly on bad code.
I’d pull something, it would be a buggy mess and I didn’t need to take apart the codebase to figure out it was terrible.
This precisely.
I consider myself a power-user. I’m going to push my system as far as j can and if I break it I either fix it or deal with it.
I also can’t code anything past “hello world” in C++ and Java, which means my ability to validate code for anything is pretty much nil. I might be able to parse a function here or there if I really dedicate to reading through every single character, but chances of my understanding the how and why of that specific function are slimmer than my chances of understanding it in the first place.
If we really want digital security, air-gapped machines running pre-2024 versions of everything is going to be the only real option. But now we’re kneecapping ourselves.
So I guess, uh, who wants to go back to Fortran cards?
Maybe there are enough objective metrics of code quality that this could be automated? It’s not just code quality though, I worry about using low effort software because maybe it will be abandoned and stop working because of broken dependencies or whatever and I’ll have wasted my time learning to use it.
Code quality automation is a big business. SonarQube, Aikido, ghql etc there’s a lot of them. No FOSS ones though. But sonarqube has a good community edition
I always use sonarqube when i vibecode. End product is so much nicer when you have some hard rules on quality. Like function complexity, duplicated code % test coverage % etc etc
This… Is a normal thing…? As in, software becoming old and outdated and replaced by something newer and better by a given metric. AIM to ICQ to Skype to Zoom to Discord to… Some of them stick around, some simply die out. As long as personal computing devices exist, you will be learning new software unless you go full Amish.
I can’t say I’m happy with the early demise of instant messaging software either (RIP msn messenger), but in that case you at least get a few years of use out of it in exchange for your time and brainspace, which may not be the case for vibecoded github projects that stop getting updates within a month of release. Of course it matters how likely it is to break too; I’d be less wary of something that is self contained and doesn’t use the internet than a third party frontend to a major website, for instance.
Here’s the real question, though, in my eyes: How do we get and keep it centered on that community? Because it hasn’t been, historically. Free Software (and all the related movements) have mostly centered two aspects that got us here.
- If you don’t like it, then just fork it.
- It’s actually wonderful that [Insert FAANG Company] exploits our labor to boost their massive profits and sometimes gives us back pizza money.
And we get this bizarre “rugged individualism” out of it, where there isn’t really any community to speak of. They’re happy to see occasional cheerleaders, but once there’s criticism, then they don’t even know who these agitators are, trying to undermine their freedoms.
If you think about it, we should have community groups checking quality (through testing and formal verification, maybe) so that it’s not on each of us to review every line of code running on our machine out of many millions. When there’s an outcry among users and rank-and-file developers that we don’t want to sacrifice quality or code provenance (or that certain people are too toxic to run things), then that should get more than a “stop being so mean to people” blog post.
The answer is “organize,” I guess, but I don’t know if that’s feasible when almost every established voice in the field insists that Freedom is something that you do on your own…
I feel we used to have a bit more organization and human aspect in the past. Back when Linux used to be more complicated and you pretty much had to talk to people to get some crazy stuff set up. I also remember going to Linux user groups, meetings… Participating on internet forums which were a bit less anonymous and transient as modern day internet…
It’s definitely possible. I always filed those groups more under the cheerleading (and technical support) column, but it’s entirely possible that (especially with much less money in the ecosystem) they exerted some power in the community. I’ve been thinking a lot about how we lost “the funnel,” recently, with even the Software Freedom Day people now strongly recommending against using events to try to sell people on Free Software…
Never mind that all the vibe coding LLMs are based on bulk harvested FOSS in the first place…
Everything i vibecode is foss. Would be pretty cringe to try to copyright or sell a product based on that.
I disclose that it is too at least.
FOSS is a bunch of licenses that use the copyright system, and whether you license your work or not, it is always copyrighted.
You get what i mean
I don’t. I also forgot that you cannot copyright (and therefore license) slop.
Well, the jury is out on that one
